The Sutton Hoo Saxon Ship – Development and Analysis of a Computer Hull Model Prior to Full Scale Reconstruction

Historic Ships, 7-8 December 2016, London, UK

THE SUTTON HOO SAXON SHIP – DEVELOPMENT AND ANALYSIS OF A COMPUTER HULL MODEL PRIOR TO FULL SCALE RECONSTRUCTION

P Handley, Paul Handley Design, UK

SUMMARY

This paper reviews recent research prior to a planned full-size reconstruction of the Saxon ship found at Sutton Hoo. The work included establishing the fundamental form of the ship by reviewing available historic information and data with consideration to possible distortion due to movement of ground and decay of timbers and likely units of measurements used based on measurement of Saxon buildings.

From this research a computer hull model was generated and hydrostatic and stability software used to assess the likely displacement, loading and stability characteristics. Performance under oar power was assessed with performance prediction software and a faired lines plan was generated from the computer model. The Woodbridge River Trust plan to commence construction of the full-size replica in 2017.

1. INTRODUCTION - A plan prepared by the Science Museum in 1939 showing the hull lines, section shape at frames and In 1939 an ancient burial mound was excavated near rivet positions for planking. A full copy of the plan is Sutton Hoo in Suffolk and the remains of an Anglo held at the Ipswich Museum and the plan is also Saxon ship uncovered. The ship was almost 27m (90ft) shown as Figure 135 in [1]. long, of wood clinker construction, propelled by up to 40 oarsmen and estimated to date to 625AD. - Information and observations recorded by Robert Bruce-Mitford and Angela Care Evans following the All of the original timber had decayed, but a well- 1967 re-excavation [1]. preserved impression of the plank and frame surfaces was left in the ground and the iron rivets remained in - Lines plans of the ship by Colin Mudie, one dated place. 1973 with hull sections at frame positions [1] and a second dated 1974, number 202.8 The excavation was cut short by the onset of war and little recorded data survives except for detailed - A plan showing planking and rivet positions, photographs of the ship impression and a provisional including a body plan with hull sections at frames, lines plan produced by the Science Museum in 1939. In (Figure 325 in [1]) described as an archaeological plan 1967 the site was re-excavated and further lines plans ‘based on 1939 photographic records and information drawn, but by then the hull impression had considerable gained by the 1967 re-excavation’. damage and the excavation team focused mainly on archeological aspects and construction. - Measurement data of the rivet positions taken in the 1967 re-excavation and a permanent record of the The available records and data were reviewed so that an 1967 shape taken via a glass fibre moulding taken to accurate computer model of the ship could be developed the British Museum to look at the stability and performance characteristics and to provide a lines plan for the planed full-scale No detailed measurement data of the shape of the reconstruction. impression in the ground in 1939 was available, nor any offset tables for the lines plans listed above. Detailed 2. HISTORIC DATA measurement data was taken in the 1967 re-excavation but by this stage the impression had been badly damaged. The recorded information, drawings and measurement data available for study for this work included: 3. OBSERVATIONS ON DATA

- Extracts from notes from the 1939 excavation written 3.1 1939 PHOTOGRAPHS by C.W. Phillips and Commander Hutchison, in Volume 1 of the Sutton Hoo Ship Burial by Robert From the photos taken the first excavation in 1939, it can Bruce-Mitford [1]. be seen that the overall shape was preserved extremely well with the rivets showing the lines of plank edges and - Detailed photographs taken in 1939, which provide a some frames showing, although there were some areas of clear indication of the overall shape of the ship missing detail and local irregularities in shape. Figure 1 impression at that time. is one of several very clear photographs taken in 1939 by

Miss B. Wagstaff. By studying the photos the following twist or angle of list, which are issues relating to the points can be observed regarding the shape of the ship: shape of the ship that were considered by the 1967 excavation team. Extreme ends - It can be seen from the photographs that no detail was uncovered at the extreme ends of the hull in The process of creating fair hull lines by hand through a 1939. Figure 2, a photograph of bow taken by C. set of unfair points would have been fairly time Phillips, shows this. The loss of the ends was probably consuming for the museum staff and they had little time due to the proximity to the ground surface and possible available to do this in 1939 due to the impending war. ploughing etc. Any drawings showing the bow or stern Accordingly the museum plan’s lines are not considered end extensions are not based on preserved evidence. to be a completely accurate record of the ship’s shape, particularly given that the ends of the ship are closer to Mid-sections - Not much detail of the shape or position the shape of the impression in the ground than to the of frames can be seen in the middle third of the length of shape that might have been expected for a ship of this the boat, but considerable detail of frame shapes was type and construction. preserved forward and aft of this middle section. Therefore, when interpreting the shape shown on any 3.3 1967 RE-EXCATVATION plans less emphasis was given to the middle sections than to the forward and aft thirds. In 1967 the Sutton Hoo ship mound was re-excavated to establish more information about the ship. Unfortunately Bow and stern sections - The photographs in Figure 2 the shape of the impression had by then been damaged shows that the bow sections of the ship impression were by military training exercises on the land during the war, flared out rather than the sharp and more vertical ends but is was still possible for the team to learn more about normally found on boats of this era, such as the Nydam the ship, in particular relating to construction and areas boat [2]. The same was true at the stern. of detail.

The general consensus since the discovery of the Sutton All rivets on the site were collected and preserved after Hoo ship is that this shape at the ends was probably their measurement positions had been recorded and caused by the planks springing away from the stem and eventually a full-scale glass-fibre casting was made of stern posts. This could be explained if the sand and soil the impression. This was made in sections to allow used to back-fill the cavity outside the hull when it was transport to be stored at the British Museum. Figure 4 buried compacted downwards over time, reducing the shows the casting assembled. support of the planks ends near the top of the mound. As the ends were near the ground surface and may have Although the casting has a boat shape, it is distorted extended beyond the surface, decay of some timbers locally and missing much of the top-sides due to the would have been faster here reducing support and damage to the impression during the war, such that both allowing the planks to move. Rupert Bruce-Mitford and Angela Care Evans stated that the re-excavation did not provide information on the Topsides - The spreading outwards of the bow and stern overall shape of the boat that could improve on that planking would have extended at least some way towards recorded by the museum plan. the middle of the boat, so throughout the length of the boat there may have been some slight outward movement 3.4 PLANS POST 1967 of the top planks as the soil settled. There seems to be some evidence of this shown in the photos. Several more provisional lines plans of the ship were drawn after the 1967 re-excavation, based on the data 3.2 SCIENCE MUSEUM 1939 PLAN available from both excavations. At least two lines plans were prepared by the Colin Mudie design office and a A plan was drawn in 1939 by the Science Museum, part further plan was produced illustrating the planking and of which is shown in Figure 3. The plan, which was some other construction detail. labelled as ‘provisional’, shows longitudinal lines and sections following fair curves without the local The section lines shown in the Science museum plan irregularities in shape that can be seen in the photos. It were compared with the later lines plans and it could be therefore seems reasonable to assume that when the plan seen that they all varied in shape. The 1973 Colin Mudie was produced, fair curves were drawn as a best fit plan showed finer sections throughout the length of the through the measurement data that was recorded at the boat than the Science Museum plan. The 1974 Mudie time. plan showed significantly fuller sections than the 1973 plan, closer to the shape shown in the Science Museum Although not stated, it is likely that measurements taken plan. Figure 5 shows a comparison of the 1973 Colin from both sides of the ship impression were averaged for Mudie body plan and the museum plan – the fine lines this exercise, thus eliminating concerns over asymmetry, are from the Colin Mudie plan.

The archaeological plan was somewhere in between the 4.2 ASSUMPTIONS FOR SHIP LINES PLAN museum plan and 1973 plan, but the hull sections were not faired, so it would seem to be more of an illustration Finer ends - For reasons describe above, an assumption of the likely appearance with some constructional detail. was made that the spoon-shape ends found in the ship impression and shown in the 1939 museum plan do not 3.5 HALF-SIZE REPLICA reflect the correct shape of the original ship. Accordingly the ends of the computer model were pulled in to create a In 1993 a working half-size replica of the ship, named shape that would be expected for planks attached to stem ‘Sae Wyfling’ was constructed by Edwin and Joyce and stern posts as typically found on ships of similar age, Gifford. The lines plans drawn by Colin Mudie were construction and style, such as the Nydam boat [2]. used for the re-construction. The replica was both rowed and sailed, but for sailing the keel timber was Bow section fullness - When the forward sections of the extended in depth to improve lateral resistance. museum plan were compared with the aft sections it could be seen that the forward sections were considerably fuller. There was no measurement data to 4. DEVELOPMENT OF HULL LINES check, but the difference in fullness could not be seen clearly in the photos. 4.1 INITIAL LINES BASED ON 1939 PLAN As a rough check measurements were taken off some In order for the 1939 Science Museum plan to be use as frame stations in the photos by measuring the camber of the reference shape for generating the initial lines of the hull segments at these frames. This exercise did not ship on the computer, copies of the bow and stern body show up much difference in curvature in the bow and sections were scanned and printed at larger scale. Some stern, so it seems like that the extra fullness in the bow missing detail at frames was added to the printed sections was unintentionally exaggerated in the museum drawings by fairing curves between points shown clearly plan, probably as a result of expediting the preliminary on the plan. lines fairing process as discussed above. The bow section fullness was therefore reduced a little in the This enhanced plan was scanned again and the points faired computer hull plan, although the forward sections representing the positions of the rivets used to join plank remain slightly larger than aft sections. edges at each frame station were digitised using Maxsurf Modeller hull design software. The digitizing was done Keel rocker - The computer hull model created to fit the by first importing the scan of the museum body plan into markers from the museum plan had a small amount of Maxsurf as a background image then inserting ‘marker’ keel rocker. This feature was retained throughout the points over the plan section lines. These digitised marker fairing process as it has been suggested by Scandinavian points were then joined automatically by Maxsurf to re- experts that a boat of this type would have had some keel create the museum body plan as a 2-D image that could rocker and it is slightly easier to generate a fair hull (or to be compared on-screen with the lines being generated for plank a hull) with a small degree of centreline rocker. the computer model of the hull. This rocker line refers to the line of intersection of the garboard planks on the centreline rather than any Maxsurf provides a number of options for automatically additional projecting keel timber. generating a 3-D hull model to fit a set of hull markers on defined stations. This approach is fine for assessing The lines generated by this fairing process show a hydrostatics and stability but as the intention was to smooth and fair overall shape with gently rounded produce a fair lines plan the new hull model was section curves running into a bow and stern shape that generated manually from a small number of control may be considered typical for the ship type. Fig 6 shows points aiming to ensure fair curves that were a good fit to the hull body section lines at frame stations with the hull the marker data. markers shown as joined crosses, showing that the shape is quite close to the museum plan at most sections A computer model of a hull was developed this way and although a fairer set of lines with slightly less curvature was a reasonably accurate fit to the marker data. If there in the bow sections. had been complete confidence with the accuracy of the Science Museum plan then this initial computer hull 5. LOAD AND STABILITY ANALYSIS model could have been presented as the lines plan for the boat. This however was not the case for reasons outlined The faired computer hull model was examined using above, so in order to develop hull lines that better Maxsurf Stability software to get a feel for the likely represent the original ship a number of further displacement, loading and stability characteristics. adjustments were made to the shape. Maxsurf allows the units to be switched from ft to metres

at any time and for this exercise metric units were used example, could all move to one side of the boat at the even though the hull form was modelled around imperial same time with only a small angle of heel resulting. measurements to reflect the likely original measurement units (see 7.1). A large angle stability assessment indicated that the hull could heel to 20 degrees at a 16 tonnes load before down- The lines plan, which is shown in Figure 7, represents the flooding, suggesting an ample reserve of stability for shape of ship with the section curves faired through the rowing but at the same time that it was unlikely that the outside surface of the planks, excluding plank overlaps. boat was intended to be sailed at an angle of heel. So if This hull shape is a close approximation to the clinker any sail was used it would most likely have been for hull shape and suitable for calculating hydrostatics, downwind sailing only. In general though, the loading, stability and other performance aspects. characteristics of the form seem to be more in line with that of a large rowing vessel rather than a sailing boat. The displacement to the estimated design waterline shown on the Science museum plan, and referred to as 6. HULL FORM AND PERFORMANCE the ‘probable light load line’, was found to be close to 16 tonnes. A rough weight estimate for the ship indicated 6.1 HULL FORM that this might represent a realistic loading for the boat, as follow: The hull model developed is fairly symmetrical fore and aft although the forward sections are slightly wider and Area of planking including overlaps is approximately fuller than the stern sections, such that the longitudinal 110 sq. m. For specific gravity of oak of 0.72 and plank centre of buoyancy to the reference waterline is 2% of thickness of 2.5 cm (1 inch) the weight of oak planking is the waterline length forward of amidships. A 1980 kg, say 2 tonnes. If frames, keel and floorboards longitudinal centre of buoyancy (LCB) forward of are taken as 2 tonnes and a further 1 tonne for thwarts, amidships is not unusual for historic ships, and given the tholes, gunwales, rivets and caulking then a rough potential advantage of greater buoyancy forward keeping estimate for the weight light-ship without ballast is 5 the bows up in rough weather or for beaching in surf, it tons. seems appropriate for the design.

If the crew number is taken at 40 with average weight of The rounded hull sections are perhaps to be expected if 100kg each including oars and some kit, then the crew the boat was designed for rowing as round sections adds another 4 tons. If 10 passengers are carried this reduce wetted surface area and hence drag, as well as might add another ton, bringing the light load achieving relatively low draught for any given load displacement to 10 tonnes. which would be helpful for operating in shallow water, such as restricted tidal areas. If stores/cargo of 3 tonnes were carried and possibly 3 tonnes of ballast, then the total loaded displacement The hull has a small amount of fore and aft rocker on the would be around 16 tonnes displacement which coincides centreline which again is perhaps to be expected for a with the estimated waterline on the Science museum rowing craft, as it both reduces the wetted surface area in plan. At this load the canoe body depth at mid section is the ends and improves manoeuvrability in restricted approximately 2 ft and midship freeboard also 2ft, which areas. Some rocker would enable the bows of the boat to might suggest that this is at the upper end of the load get closer to the shore edge when running up a beach to range rather than a ‘light load’ as stated on the museum enable crew to disembark without getting too wet. plan. Further, a small amount of keel rocker would enable the boat to be freed and turned more easily after grounding. The hydrostatics were run at a series of waterlines and So unless there was a need for deeper ends and keel the load range from 10 to 16 tons looked appropriate for depth - for lateral resistance for sailing - it seems the design. As an absolute limit, the displacement when reasonable that these features would be absent and the loaded to the point that the gunwales were at the hull would have a small amount of rocker with shallow waterline was found to be 50 tonnes. Given the very keel timber. light construction of ship it seems unlikely that it would have been used for heavy load carrying and this also 6.2 POWERING AND RESISTANCE raised the question as to whether or not any ballast would have been carried if it was solely powered by oars, as The prismatic coefficient for the model was light weight would have been beneficial for performance. approximately 0.6. This is towards the high end suggesting a hull capable of reaching a displacement Stability was checked for the range of load cases from 10 speed of around 9 to 10 knots for a hull of this length and to 16 tonnes based on estimated vertical COG for the relatively low power from rowing. loads listed above, with no cargo load or ballast included for the 10 tonne load case. At each of these load cases The performance of the hull with respect to resistance, the stability was substantial, such that ten persons, for power and likely speed was further assessed for the final

lines using both empirical and calculation methods in It therefore seems a reasonable assumption that the Maxsurf Resistance software. Power estimates for a Roman/current imperial measurement system was used range of speeds were calculated at 13 tonnes for the Sutton Hoo ship, particularly when further noting displacement by three empirical methods (Holtrop, Fung that the overall distance between the end-most frames is and upright resistance from the Delft III sailing VPP) 75 feet or 5 ‘rods’. and by applying slender body theory, which is applicable for the length to beam ratio and displacement of the ship. Also, that the average planking thickness is 1 inch based It can be seen from Figure 8, which shows the predicted on measurements taken from plank scarf join rivets. And power curves, that the methods compare quite closely while it could be coincidental, it was noted that the over the expected range of speeds. The approximate average width of the hull planks in the mid sections is predicted power is 10hp at 8 knots, 20hp at 9 knots and approximately 1 ft - see 7.2 below. 30hp at 10 knots. If this measurement system assumption is correct, then it It is difficult to estimate the power delivered by a human would suggest that there has been no significant with an oar, and also there is a question over the number expansion or contraction of the ground over the 75 ft of oarsmen, which may have been 28 or 40, depending length between the end frames, and that the overall on whether or not tholes for oars were originally fitted in dimensions of the impression may be considered very the mid section of the ship but removed for the burial. accurate, perhaps more so than the dimensions of preserved historic wooden ships where timber distortion 6.3 SEA-KEEPING AND STRUCTURAL and shrinkage can lead to considerable speculation over INTEGRITY the original ship dimensions and shape, as is the case for the Nydam ship, as shown in Figure 10. As a further exercise the computer hull model was animated in waves of 1m height and 2m height. In 7.2 PLANKING AND FRAMING waves of 1m it could be seen that the freeboard remained almost uniform as the bow and stern lifted the hull to the The plank runs as recorded by the rivet positions shown waveform. At 2m the ship appeared to be taking some on the Science Museum plan were represented on the water aboard at the ends, which could have been computer by marker positions. There were some deflected by temporary decking. However, it seems discrepancies found in the shape of the longitudinal lines unlikely that the ship would have operated in large waves passing through these points, which represent the plank as it’s lightweight construction and low amidships overlap paths, particularly in the mid portion of the boat, freeboard may well have lead to structural issues. which is perhaps to be expected due to the lack of detail uncovered in the mid section. As a further exercise the structural strength of the ship could be assessed to establish what the maximum load The plank widths (from plank edge rivet to rivet) were and wave height combinations would be whist still recorded at longitudinal positions during the re- maintaining a good margin for structural integrity. excavation and although this information is also mostly missing for the mid portion of the boat it does give a good idea of the average plank width, which is around 1 7. CONSTRUCTION ft, excluding plank taper in the boat ends. This is based on 10.5 inch average distance between the edge rivets 7.1 SCANTLINGS AND MEASUREMENTS plus 1.5 inch plank overlap.

Measurements of the frame positions taken from the As a further check on plank width the girth of the plans showed that the average frame spacing, ignoring midship frame section of the computer hull model was irregular spacing at the ends of the boat, was very close measured and it was found that the correct girth would be to 3 feet (2.994 ft average). achieved by 9 clinker planks each of 1ft width with overlaps around 1.3 inches. This was based on a half- This 3-foot frame spacing may be coincidental, but a girth measurement of 8.5 ft less a small amount for the review of measurements used in Saxon times for land keel timber width and a slightly wider sheer plank. buildings indicated that it was likely that there were two measurement systems in use at the time. The plank timber is thought to be oak. The planking would have originally been made by splitting trees and One was based on a ‘rod’, which was 15 German feet trimming the planks with adze or axe [3]. The intention long with each German foot 13.2 inches long, so the is to attempt to produce planks in the same way for the German rod was 16.5 modern feet long. The other was reconstruction of the ship. based on Roman units where a ‘rod’ was 15 feet long with each foot 12 inches long, these feet and inches being The lines plan derived from the computer model shows a the same size as modern imperial units. frame spacing of 3ft except for the end frames beyond

the drawn waterline, so that the frame positions will 10. REFERENCES coincide closely with the original frame positions. A nominal design waterline was drawn with a draught of 1. BRUCE-MITFORD, R., Volume 1 of The 2ft which gives a waterline length of 60ft with the Sutton Hoo Ship Burial, British Museum, 1975 forward perpendicular (FP) at Frame 3. 2. ABEGG-WIGG, A., ‘Das Nydamboot’, 2014 8. CONCLUSIONS 3. CONSTANTINE, P., ‘Anglo-Saxon Craft Despite the complete absence of any of the original Construction’, 2015 timbers, it can be concluded that the impression in the ground provided a remarkably good record of the ship’s 11. AUTHORS BIOGRAPHY shape, perhaps even more reliable than historic ships that are found intact but subject to indeterminate distortion Paul Handley MRINA is an independent consultant and shrinkage of the timbers. Such was the preservation Naval Architect and small boat designer. He was a of the layer between the timbers and surrounding sand consultant to the European Standardisation Committee and soil, and the ‘map’ of rivets, that it was also possible for eleven years assisting with development of ISO to establish considerable construction detail too. standards supporting the Recreational Craft Directive and continues to provide consultancy on boat safety standards Taking into consideration the available recorded data on and design. His boat designs include two ISAF the shape of the impression and subsequent analysis, a international youth sailing classes, the RS Feva and RS fair computer model of the hull has been generated. In Tera, several adult dinghy and keelboat classes and addition to providing a lines plan for the planned rowing sculls. reconstruction of the ship this model provided useful information on stability and likely performance.

The Woodbridge Trust plans to start construction of the full-scale replica in 2007 at a special site in Woodbridge, where the boat will eventually be based for public viewing.

9. ACKNOWLEDGEMENTS

The Author would like to thank the Woodbridge Riverside Trust and in particular Paul Constantine for proposing the project to create a computer model of the Sutton Hoo ship and for providing regular input to the project. Thanks are also due to Peter Bradbeer and Joe Startin for their enthusiastic input.

Figure 1: The ship trench in 1939, photograph by Miss B. Wagstaff

Figure 2: Bow sections of the ship impression, photograph by C. Philips

Figure 3: 1939 Science Museum plan

Figure 4: Casting of the 1967 ship trench, photograph by BBC

Figure 5: Comparison of body plan by Science Museum and Colin Mudie (1973)

Figure 6: Computer body plan lines compared to Science Museum lines (through makers)

Figure 7: Lines plan from computer model

Figure 8: Predicted power curves for Saxon ship

Figure 9: Ship in 1m waves

Figure 10: Interpretations of shape of Nydam boat